INTERNATIONAL COUNCIL FOR BUILDING RESEARCH STUDIES AND DOCUMENTATION WORKING COMMISSION W18- TIMBER STRUCTURES

Transcription

1 CIB-W18/ INTERNATIONAL COUNCIL FOR BUILDING RESEARCH STUDIES AND DOCUMENTATION WORKING COMMISSION W18- TIMBER STRUCTURES MULTIPLE-FASTENER DOWEL-TYPE JOINTS, A SELECTED REVIEW OF RESEARCH AND CODES by C J Mettem AV Page TRADA Technology Limited United Kingdom MEETING TWENTY- FIVE AHUS SWEDEN AUGUST 1992

2 MULTIPLE-FASTENER DOWEL-TYPE JOINTS, A SELECTED REVIEW OF RESEARCH AND CODES C.J. Mettem and A.V. Page TRADA Technology Limited Introduction It is frequently observed that the load-carrying capacity of a multiple-fastener joint is less tban the sum of the individual fastener capacities. For dowel-type fasteners, such as plain round metal dowels, bolts, and nails, a number of thnber codes now take some account of this effect. As already noted in CIB W18 papers by Blass (1) in 1990, and previously Steck m 1986 (2), there are great differences between the modification factors for the number of fasteners, in several international and national thnber codes. This still seems to be the case in A paper accompanying this brief review, also to be presented at the current meeting, reports recent tests supported by a linear elastic finite element analysis. The work was conducted on both single and multiple bolted joints in European whitewood glued laminated thnber (glulam) (3). Strain gauged steel side plates were used, following an established technique adopted by previous researchers. By this means the load taken by each individual bolt in the multiple-bolted joints could be measured. However, the ratio of fastener length to central thnber thickness was kept very low, so that pure embedment testing was achieved. This was new. Parallel and perpendicular to grain loading arrangements were included. A substantial lack of "evenness" in load carrying capacity was demonstrated, both by test and by theory. This applied especially in the perpendicular to grain case, where the study suggested that modification factors as conservative as 0.4 should be considered for multiple-fastener joints including as few as four bolts. If this is correct, then it is evident that the current draft of EC5 is unsafe in this respect. Background The multiple-fastener load reduction effect is also known to exist in materials other tban thnber. For example Volkersen analysed rivet stress distributions in aircraft structures, using as an analogy theories originally developed for adhesive stresses (4). Care is of course necessary when drawing analogies between one subject of engineering science and another. However, the extensive research on uneven stress distributions in glued joints, which can be formed using a lap, in single or double shear, is in this case pertinent. This has been reviewed in previous TRADA research (5). In the stress analysis of glued joints, it is found that the unevenness of the stress distribution along the joint becomes more pronounced, as the discrepancy increases between the stiffness of the adhesive and that of the adherend (6). Conversely, research on multiple-fastener joints in timber has not always revealed a major unevenness. For example, Blass reported that in multiple-nailed double-shear joints loaded parallel to the grain, the maximum load of a multiple fastener joint can be estimated as the sum of the values for individual nails, provided joint failure is by nail yielding.

3 2

4 However, this difference is not necessarily inconsistent. The nailed joints which Blass investigated, in which the stiffnesses of the nails and the timber are comparable, are analogous to glued joints made with an adhesive which has a stiffness comparable to that of the timber, the very joints in which the stress distribution is most even. Yasumura et a!, on the other hand, conducted a series of tests on multiple numbers of bolts and multiple rows of the same, connecting glulam members to steel side plates, the latter forming double shear planes (7). They found that for certain parallel to grain configurations, with a reasonably practical number of bolts and rows, the average load per bolt in the multiple configuration was only about half the value for the specimen having a single bolt. Design Situations Since the constituent parts of mechanicaily fastened timber joints often have very different stiffnesses, the above considerations suggest that practical situations do exist where the load distribution amongst the individual fasteners is markedly uneven. One such situation is where relatively thick metal plates, usuaily mild steel, are used in conjunction with bolts or plain round dowels, to form connexions in glulam. This is a common form of construction. When a line of fasteners perpendicular to the grain is loaded in that direction, the unevenness in the distribution of load is generally even more pronounced, because the stiffness of timber perpendicular to the grain is only about one sixteenth of its stiffness parallel to the grain. This means that the actual strength of the joint is considerably less than the strength predicted from tests on single bolts loaded perpendicular to the grain, reduced by a factor for the number of bolts in a line based on tests parallel to the grain. The reallty of this effect was measured and confirmed in tests conducted by TRADA. Furthermore, the situation is quite common in practice, as shown in the accompanying diagrams, Figure 1. Eurocode 5 and BS 5268 Draft Eurocode No. 5 'Design of Timber Structures', Part 1, April 1992, contains a principle under Section 6, JOINTS, that: P(2) It shall be taken into account that the load-carrying capacity of a multiple-fastener joint will frequently be less than the sum of the individual fastener capacities. However, the modifications for the effect, given in the application rules, are remarkably modest. As is by now well-known, EC5 recommends to the designer that the lateral load-carrying capacity of doweltype fasteners should be calculated from a set of formulae based on a theory initiated by Johansen (8). Under the relevant clause for steel-to-timber joints, a sub-set of these formulae is given, together with small explanatory diagrams. These give load carrying capacities per fastener per shear plane. Under Section 6.5, BOLTED JOINTS, EC5 advises that for more than six bolts in line with the load direction, the load carrying capacity for the extra bolts should be reduced by one third. Taking as an example nine bolts in line, this equates to a "modification factor", as BS 5268 would call it, of only The British code itself would apply a modification factor of 0.76 for this case. Dowelled joints are, by an "extension" clause, treated by EC5 in the same manner as bolted joints in this respect. Neither EC5 nor the British code identify as a special case the situation where the timber is loaded perpendicular to the grain, or where there is a component of load in this direction. 3

5 North America In North American codes, the larger and stiffer types of standard fastener, such as bolts, lag screws, split rings and shear plates, are treated differently to smaller and less stiff types such as glulam rivets and nails. For the latter, the Canadian code, for example, CSA M89 (9) has no modification factor for the number of fasteners in the group. For the larger types on the other hand, the overall reductions for the group effect often produce more conservative rules in practical design cases than the rules of draft EC5 and BS Recent proposals for North American codes also give different modification factors for loading parallel and perpendicular to the grain. CSA-086 gives a modification factor J G which varies according to the number of fasteners in a row. For connectors and lag screws, this is presented in tabular form, the value being selected according to an "area ratio" which takes account of the cross-sections of the main and side members. For steel side plates, this factor can range from 0.42 to Bolt design, according to the Canadian code, involves the use of two pertinent modification factors. These are J G, which has the same purpose as above but which is computed by means of a formula, and J R, which is a factor for the number of rows, ranging from 1.0 for one row to 0.6 for three rows. A general clause tends to preclude the use of wide steel splice plates altogether, thus avoiding the situation of numerous rows of bolts. It is understood that the tests leading to the "number of bolts" formula were made at Carlton University, on behalf of 'Agriculture Canada'. The J R factor was derived from the work reported to CIB by Yasumura et al. In the USA, the approach used in the National Design Specification 1986 (10) is at present similar to that described for Canada, with the "Group Action Factor", C g, at present depending upon the number of fasteners in a row, and upon the main member to side member area ratio. This procedure is used for all the "large" types of fastener, including bolts. In the past, research in the USA has shown the need for quite substantial reduction factors for group action, although results have not always been consistent. Doyle, for example, (11) examined joints made with four bolts in two rows, in Douglas fir glulam, using steel side plates. For parallel to grain loading, he derived reduction factors from approximately 0.6 to 0.9. Kunesh and Johnson (12) investigated similar multiple-bolt joints and obtained factors ranging from approximately 0.5 to Wilkinson (13) conducted extensive tests on a range of numbers of bolts. He also developed an analytical method by modifying the earlier work of Cramer (14). Wilkinson's experimental method was to use strain gauged steel side plates as a means of measuring the load carried by each bolt. However, his main timber members were of a thickness such as would be used in real construction, rather than thinner timbers where the l!d ratio would be low, producing a pure embedment effect. Furthermore, he reported "fabrication effects" in the manufacture of the specimens, such that these had a greater influence on the distribution of load amongst the bolts than the properties of the members and the fasteners themselves. Amongst Wilkinson's conclusions were the observation that the load distribution for any particular row of bolts is unique. Any one of the bolts may be the major load carrier. Also, any bolt hole may be misdrilled, causing that bolt to carry almost no load for a major portion of the joint loading. 4

6 Proposals for a 'Load and Resistance Factor Design' code in the USA include in the draft a more sophisticated multiple fastener modification factor. Again, this is to apply to all "large" fasteners, namely bolts, Jag screws, dowels, shear plates and split rings. It is understood that the theory has been worked by Zahn, following linear elastic analytical predictions given previously by Lantos (15) and Cramer. Both row factors and factors for fasteners within a row are included in a single formula in the LRFD draft. This also takes into account the axial stiffness (EA) of the main and side members, as well as the load/slip constants for the fastener concerned. Whilst leading to a formidable formula to be computed by the designer, this all-embracing approach would take account of effects such as the large differences in stiffness of the constituents of a joint involving a member loaded perpendicular to the grain. At the same time, no major difference in approach has been adopted. Thus, for example, fracture mechanics considerations and the transverse tensile properties of the timber are not taken into account. Further Research? It was not the intention that this review should be long or comprehensive. It is understood that the topic of multiple-fastener joints is also being re-considered in other countries, such as New Zealand. It is apparent, however, that the statement given at previous CIB-WlB meetings that there are considerable differences between the modification factors for number of fasteners amongst various codes is true. matter. Whether this indicates that research on the subject should continue indefmitely is another The topic seems in danger of becoming a timber researchers' "chestnut". It is capable of being treated with an almost infinite amount of analysis and testing. Consider, for example, the following influences and aspects which it might be considered necessary to examine, listed in no special order of significance: Brittle fracture modes of failure Mode changes with time in service, due for example to creep deformations and increased embedment Re-loading situations Moisture movement effects, especially where timber is restrained by steel Fabrication effects such as hole alignment and mis-fit; tightness of fasteners in holes For perpendicular to grain loading, the substantial influence of crack propagstion and weakness in the timber Load-slip behaviour beyond the elastic range Code Implications Code writers cannot await indefmitely the results of further research, and must make judgements on the basis of present knowledge. It can also be argued that unless there has been evidence of failure or unsatisfactory service, then there is not a pressing need to alter codes or to make design procedures more complicated. However, there have been instances of failure in multiple fastener joints using the "large" types of device discussed here. 5

7 It is difficult to pinpoint the primary causes of such failures, since "real" joints in built structures experience a variety of the effects described in the "research menu" given above. For example, fissures induced by moisture content changes, in combination with perpendicular to grain loading, seem to be a particularly serious hazard. It is perhaps especially disconcerting, therefore, that the European codes mentioned in this review take such a sanguine view of the perpendicular to grain loading case for multiple fasteners, not singling it out for especially cautious treatment. It also seems odd that the latest draft of EC5 still treats ail dowel-type fasteners in the same manner with respect to multiple-fastener modification effects, whereas other codes, including those used in North America, fmd no modification necessary for nails, but recommend substantial modifications for the "large" types of fastener such as bolts, lag screws, split rings and shear plates. Multiple-fastener dowel-type joints using bolts or plain metal round dowels in conjunction with steel plates need careful detailing and EC5 and BS 5268 at present carry insufficient warning as to the hazards of their use. It is also likely that more conservative application rules should be given, to fulfil the principle that the load-carrying capacity will be less than the sum of the individual fastener capacities. Perhaps those responsible for the final re-drafting of EC5, prior to the issue of the EN version, might wish to consider revising the reduction factor recommended in Clause (3) for the number of fasteners in a line, even if the researchers' thirst for further knowledge has still not been fully slaked. 6

8 References [1] Blass, H.J. Load distribution in nailed joints. CIB-W18N23-7-2, Lisbon, 1990 [2] Steck, G. Effectiveness of multiple fastener joints according to National Codes and Eurocode 5 (Draft). CIB-W18N19-7-3, Florence, 1986 [3] Mettem, C.J. and Page, A. V. Load distributions in multiple-fastener bolted joints in European whitewood glulam, with steel side plates. CIB-W18/ , Alms, 1992 [4] Volkersen, 0. Die Nietkraftverteilung in zugbeanspruchten Nietverbindungen mit konstanten Laschenquerschnitten (Rivet stress distribution in tensile stressed rivet joints with constant butt strap cross-section), Luftfahrtforschung, Band 15, pp , 1938 [5] Page, A. V. Behaviour of structural glued joints in wood. Research Report RD007, TRADA, 1989 [6] Adams, R.D. and Wake, W.C. Structural adhesive joints in engineering. Elsevier App.Sci.Pub., 1984 [7] Yasumura, M., Murota, T. and Sakai, H. Ultimate properties of bolted joints in gluedlaminated timber. CIB-W18N , Dublin, 1987 [8] Johansen, K.W. Theory of timber connectors. IABSE, Publn.9, Bern, 1949 [9] Canadian Standards Association. Engineering design in wood (Limit States Design). CAN/CSA M89, Toronto, 1989 [10] National Forest Products Association. National design specification for wood construction. Washington, DC, 1986 [11] Doyle, C. V. Performance of joints with eight bolts in laminated Douglas fir. USDA Forest Service Research Paper FPL 10, US Govermnent Printing Office, Washington DC, 1964 [12] Kunesh, R.H. and Johnson, J.W. Strength of multiple-bolt joints; influence of spacing and other variables. Oregon State University, Forest Research Laboratory Report T-24, Corvallis, 1968 [13] Wilkinson, T.L. Load distribution among bolts parallel to load. Journal of Structural Engineering 112(4): , USA, 1986 [14] Cramer, C.O. Load distribution in multiple-bolt tension joints, Jnl. of the Struct. Div., ASCE, Vol. 94 No. ST5, Proc. Paper 5939, pp , USA, 1968 [15] Lantos, G. Load distribution in a row of fasteners subjected to lateral load, Wood Sci. Vol. 1, No. 3, pp , UK,

9 ., Multiple-fastener dowel-type joints with steel plates a) l c) ,?) 1,9) \_9) -;3 { I "" " t'... /... ' l,? )!0' J ' ro 1,_ /' {0\,, ' '>I f 0\,,., b) { I I { d) a) to d) Loading parallel to grain. e) D e), D Loading perpendicular to grain. g) h) g), h) Complex loading. 8